J. Soc. CosmeticChemists,19, 649-667 (Sept.

16, 1968)

Color and Its Measurements*

MICHAEL BORNSTEIN, Ph.D. t

Synopsis--An account is given for the fundamental principles of diffuse and tristimulus re-
flectancemeasurements. The experimental basisfor these measurements,including the CIE
and Munsell systemsfor defining color space, is also discussed. These principles are sup-
ported with qualitative and quantitative examples for color matching, as well as interaction
studies with reflectance values and the Kubelka-Munk equation. A review of several availa-
ble commercial color measuring devices, reference standards, and sample preparation tech-
niques are also presented.

INTRODUCTION

The stability and appearanceof colored cosmeticand pharma-

ceutical dosageforms are dependent on the dye or lake employed, as
well as many other parameters. These may include adjuvant absorp-
tion coefficientsof microscopic and macroscopic structures as well as
otherphysicaland chemicalproperties(1). Althoughmostformulators
still rely on empirical knowledgein the subject of color measurement,
many articles have appeared with respect to color instrumentation.
Swartz and Cooper (2) have reviewed a number of pharmaceutical
colorantsand their properties. Lachman and associateshave applied
various parameters to color stability of tablet formulations (3-5).
Goodhart, Lieberman, and associates(6, 7) have alsostudiedthe stabil-
ity of certified dyes in tablets. Raft (8, 9) has presentedreports of
color measurements obtained from FD&C colorants.
The purposeof this communicationis to discussprinciplesand instru-
ments used in diffuse and tristimulus reflectance measurements and their
applicationto the developmentof coloredpharmaceuticaldosageforms.

* Presented October 2, 1967, Sixth Annual A. Ph. A. Industrial Pharmaceutical Techno-

logy Meeting, Chicago
p Pitman-Moore Division, Dow Chemical Co., Research Center, Zionsville, Ind. 46077.
Present address, Eli Lilly & Co., Indianapolis, Ind. 46206.
649
650 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

CONCEPTS OF COLOR

The sensationof coloris the result of radiant energy, between400 and

700mu, impingingthe eye (10). Sincethe phenomenonof colorincludes
psychological as well as physicalaspects,color scalesand instruments
are often designedto include both parameters. The psychologicalas-
pectsof colorare the resultof colorcodesstoredin the brain, including
associationof an individual hue suchas blue-greenwith a characteristic
sensationsuchascoolness. Physicalaspectsof colorresult from the fact
that visible radiant energyis necessaryfor vision.
The basicprinciple on which tristimulus color measurementis based
dependson the assumption
that mostcolorsmay be producedby a com-
bination of red, blue, and green coloredlights. Red and green beams
combine to form yellow, whereas blue and green combine to form a
bluish-greencolor. A purple colorresultsfrom the combinationof red
and blue; a white hue forms when the red, green, and blue primaries
unite. By varying the amountsin the three primary beams,all inter-
mediate colors can be produced. This phenomenonof uniting three
basic colors to form white is the basis for an additive color mixture.
Onemay alsoform a subtractivecolormixture by usingyellow,magenta,
and cyan pigments. The resultant color darkenswith the subtractive
mixture as more of thesefilters combine; when all three subtractivepri-
maries unite, a black color is formed. Numerous systems have been
developedto definecolor. This discussionwill be primarily concerned
with the CIE and Munsell systems.
CIE SYSTEM

There are two basicfoundationson which colorimetrydepends. The

first is that color can be matched by a suitablemixture of three selected
light radiationsand that if two colorsare matchedby three radiations,
the mixture of these two colors is found additive by suitable optical
means. The essential principles of this tristimulus color system were
first independentlydevelopedby Ives (ll) andGuild (12). In 1931,the
CommissionInternationale de l'Eclairge (CIE) standardizedthe color
mixture characteristicsof an "average observer" and developeda stand-
ard framework for a color specification.
This standard observer represents a series of functions determined
from data providedby observersmatchingthe color at eachwavelength
from 400 to 700 m•, with mixturesfrom three primary light sources.
The experimentalset-up for definingthe "average" human observer
measuresthe observer'sresponseto the three primary colorsby focusing
 COLOR AND ITS MEASUREMENT 651

SPECTRUM COLORS

1.6
• GREEN• 'x 540

• 1.2
søø

• • RED
I BLUE
• [ 4•o •.'c' •• ,oo

0
400 500 600 700 400
• m•u

Figure 1. Responseof the "standard CIE Figure 2. Chromaticity diagram illustrating

observer" to an equal energy light source the visible spectrum locus measured with 10
spectrophotometric readings. The tristimu-
lus chromaticity coordinates (x,y) for illumi-
nant C are plotted near the center of the chart

the threelightson a screenin the samelocationsothat they may be mixed

in properproportions. A monochromaticlight sourceis focusedon a spot
just adjacentto the mixture of the three coloredlights. The observer
views the screenthrough a cone angle of two degreesand is asked to
adjust the three primary colorsuntil the mixture matchesthe mono-
chromaticlight source,recordingthe relative amountsof the three lights
necessaryfor the match. Figure 1 showstheserelative amounts of red,
green,and blue lightsbetween400 and 700 mu neededby the observer
for the match. The monochromaticlight is ideally a "white light,"
meaningthat it is an equal energysourcecontaininglight from all visible
wavelengths. Most standard light sourcesdo contain light of every
color, but not in equal amounts. Illuminant A, found in standard in-
candescentor tungstenlight sources,has light at all wavelengths,with
muchmore energyat longerwavelengths. Illuminant B is a light source
equivalentto the noonsun. Illuminant C, representingdaylight on the
north side of a building, contains light of all wavelengths with much
more energyin the blue part of the spectrum. Figure 2 showsthe three
dimensionsof the CIE color solid in two dimensionsusing x and y chro-
marlcity coordinates. These chromaticity coordinatesare calculatedby
expressingthe tristimulus values X, Y, Z, as fractions of their total
These tristimulus values of a sample are either calculated from diffuse
reflectancemeasurements(9) to be discussedlater or they may be mea-
652 JOURNAL OF THE SOCIETY OF COSMETI•2 CHEMISTS

• •DOMINAN
ILL0
'C'
½• X

F{gur½ 3. Chromaticity diagram (x,y) defining

i]]um{nant C, dominant wavelength, and purity

sureddirectlyontristimuluseolorimeters. Only two coordinates needto

be specifiedsincethe total of the x, y, and z chromaticitycoordinates
equalsunity. Here the coloraxisor chromaticitycoordinates lie on the
outsideof the triangle goingfrom blue to greento red, whereasthe white
to black axis or luminosity function is the center; this brightnessaxis
wouldbe perpendicular to the screenwith the white at the top andblack
at the bottom. Nonspectrum colors representmixtures having the
chromaticitiesrepresentedat any point alongthe straightline joiningthe
extremitiesof the spectrumlocusand are producedby mixing suitable
portions
ofradiantengery
takenfromtheextreme
short-wave
(less
than
440 m/•) and long-wave(greaterthan 680 m/•) regionsof the spectrum.
Colorsrepresentedby pointson straight lines betweenthe achromatic
pointC in Fig. 2 andthe spectrum locusarecalledspectralcolors.
The CIE systemdividesthe characteristics of light from whichcolor
is composedinto severalcomponents includingluminance,dominant
wavdength, and purity. Luminanceis the characteristicthat differ-
entiatesthe light reflectedfrom a standard white sample,illuminated
with a 100-W lamp, from that of light reflectedfrom this samewhite
samplewhen it is illuminatedby a 200-W lamp with all other things
being equal. The dominantwavelength may be definedas the wave-
lengththat appears to bedominantin the lightandisusuallythemostin-
tensevariety of radiant energyin the stimulus. Purity refersto the de-
greeto whichthe dominantwavelengthappearsto predominate in the
light.
Figure3 showsthat the dominantwavelength is obtainedby plotting
the chromaticitycoordinatesof the illuminant and sample. A line is
 COLOR AND ITS MEASUREMENT 653

White

8
9
IO
1
2
3

.10

Black

Neutral • Saturated
colour Chroma colour

Figure Hue, value, and chroma coordinates of the Munsell system

drawn from the achromatic or illuminant point through the sample co-
ordinates onto the locus of the diagram. This point on the locus is
calledthe dominant wavelength. The per cent purity is determinedby
the ratio of the distancefrom the illuminant point to the samplecoordi-
nates, A, divided by the distancefrom the illuminant point to the spec-
trum locus,B, expressedin per cent. Luminosity is expressedin per
cent Y, obtained directly from tristimulus colorimeters. This diagram
in which each point representsthe chromaticity, independentof lumi-
nance,
ism,
called
achromaticity
diagram.
MUNSELL SYSTEM

Another systemfor definingcolor space,called the Munsell system

(13), is basedon a colorsolidillustrated in Fig. 4. The central vertical
axisrepresentsthe locusof neutral colorswith white at the top and black
at the bottom. In this system, lightnessof the sample, called value, is
divided into a number of stepsfrom 0 to 10. The distanceof the sample
from the central vertical axis is a function of the saturation or intensity
654 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

of color and is called chrornaon the Munsell map. Chroma is also

divided into a seriesof steps, with the neutral axis or gray being 0,
whereasa fully saturatedsamplehasa chromaof 12. Color, or hueasit
is called in the Munsell system (14), is presentedby different planes
around the vertical axis of the Munsell solid. In this system,the whole
solid is divided into ten equal vertical segmentswith five principal hues
(red, yellow, green, blue, and purple) occupyingthe central planesof
alternate segmentswhile intermediate hues (yellow-red, green-yellow,
blue-green,purple-blue and red-purple) occupy the remaining planes.
Each hue segmentis further divided into ten sectionsnumbered 1 to 10,
with the main hue segmentalways numbering5.
The hue of a sampleis designatedby a number, indicating the section
of the segment,followedby symbolswhich showthe colorof the segment
involved. For example, 10P indicatesthe hue section10 of the purple
segment. To completethe specificationof the sample,the value quan-
tity followsthe hue which is then followedby a stroke and then the
chroma. Here the colordefinedat 10 P •/• hasthe hue 10 P, the value5,
and the chroma of 8.

COLOR DIFFERENCE AND TOLERANCE

It is of interest to point out that color differencesare not entirely

linear throughout the entire CIE chromaticity diagram since,if one cal-
culatestwo points on this diagram, another set of points twice as far
apart do not necessarilyhave a totally linear differencein color (9, 15).
In other words, equal distanceson the diagram do not always indicate
the samedegreeof color change.
Judd (16) devisedthe Uniform ChromaticityScale(UCS) adoptedby
the National Bureau of Standards(NBS). Adams (17) followedwith his
chromatic-valuediagram which was acceptedby the American Society
for Testing Materials (ASTM). MacAdam (18) investigatedcolordif-
ferenceswhere the observermade a large number of matches on a series
of test colors located at various points in the chromaticity diagram.
The spreadin the colorsettingsnecessaryto make a match (Fig. 5) was
used by MacAdam as the criterion of sensitivity to color differences.
The distancesfrom points on the boundary of each ellipseto the central
point within the ellipseapproximatelycorrespondto the chromaticity
differencejust perceptiblewith certainty under specifiedviewing condi-
tions; this just perceptible color differenceis defined as one MacAdam
unit. The orientation of the major axis of individual ellipsesis a func-
tion of the dominant wavelength and purity changes, but does not
 COLOR AND ITS MEASUREMENT 655

Y 1.s
52.• 530
•0 •'X •4•

o.4-499 • • •oo

F•gure5. MacAdam's perceptiblechromaticitydifferencesin the CIE diagram.

The axis of each ellipse has been multiplied by 10 for illustrative purposes

MONOCHROMATOR

SOURCE
MPLE

PHOTO-SENSOR
Figure6. Schematicdiagramrepresentingthe major componentsof a spectrophotometeror
colorimeter. Spectrophotometers use prismatic or grating monochrometers where colori-
meters would contain filters

changewith lightness; i.e., the shape and orientation remain constant

at differentluminosities. The ellipsesdo vary in size with a change
of lightness,becomingsmaller as lightnessincreases,since errors in
makingcolordecisionsdecrease underimprovedlight-viewingconditions.
COLOR MEASURING INSTRUMENTS

Reflectance
Spectrophotometers
This type of instrument,diagrammedin Fig. 6, is basedon a beam of
monochromatic light penetratinginto a sample. This light is scattered
in many directions,is partially absorbed,and finally re-emerges to the
656 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

400 700 400 700

SAMPLE REFLECTANCE (R) SOURCEEMISSION(E)

RxE

SPECIFYING THE COLOR OF A SAMPLE

Figure 7. Schematic diagram for interaction

factors to produce color vision

surface. Measurementsmay be plotted as reflectancers. wavelength

using a relatively nonabsorbingwhite referencestandard. This infor-
mation is then usedfor quantitative Kubelka-Munk equationsor it may
be converted to tristimulus values which will be discussed.
Commercial instruments include the Hardy General Electric Spec-
trophotometer,Bausch& Lomb Spectronic505 with reflectanceattach-
ment, as well as the Beckman DU Spectrophotometerwith the reflect-
anceaccessory. The G. E. instrument has a recorderand automatically
converts reflectance data to tristimulus values. The Bausch & Lomb
instrument has a recorder whereas the Beckman DU is a manual spec-
trophotometer.
Tristimulus Instruments

It should be pointed out that the CIE system was developedto

eliminateeye variationsby usingthe "standardobserver." The princi-
ple behind tristimulus colorimeters,illustrated in Fig. 7, is as follows:
A light source,E, strikesa samplehavinga reflectancecurve,R. This
interaction resultsin a reflectanceenergy,RE, which is specificfor each
wavelength; this is alsothe energy striking the eye. This resultingRE
passesthe filtersand is multiplied by eachof the hypotheticalcolormix-
ture curvesœ,9, and •, representingcolorvisionof the standardobserver
in the CIE system. This producesthree new curves,each having an
area representedby the tristimulus values X, Y, and Z. The above
manipulationsagain indicate that the CIE systemspecifiesa color by
three quantities,X, Y, and Z, calledtristimulusvalues. These values
 COLOR AND ITS MEASUREMENT 657

representthe amounts of primary red, green, and blue color that the
standard observer would need to get a match. Ii each of these tri-
stimulusvaluesis divided by the sum of three, the resulting values x, y,
and z, called chromaticity coordinates,give the proportion of the total
stimulus attributed to each primary color. Furthermore, since the
sum of the three chromaticity coordinatesis unity, the valuesof x and 3'
plotted on the chromaticity diagram may be used alone to specify the
color. The third achromatic dimension of lightness or darkness is
specifiedby the ¾ tristimulus value. The manipulations included in
tristimuluscolorimetryare mathematically presentedbelow:
Weighted Ordinate Calculation of
Method Chromaticity Coordinates
700 X
X = • ExC'Rx.•X x=
•oo X + Y + Z
700 y
Y = •oo
• 'Exc'RX':?x y=X + Y + Z
700 a
Z = • Exc'RX.zX z=
•oo X + Y + Z

z = I-- (x-t-y)
The Hardy General Electric Spectrophotometer,
previously dis-
cussed,provides spectral reflectancecurves in addition to tristimulus
values. This combination results in nonmetameric matches in color
formulationswhich are colorssimilar in appearanceunder all light
sources. Tristimulus instruments alone often produce metameric
matcheswhichlookthe sameonlyif similarviewingconditionsare used.
Instrument Development Laboratories(IDL) market the Color-Eye
whichis an abridgedspectrophotometer giving10 or 16 visiblewave-
length points as well as tristimulus values. The instrument is also de-
signedto measurefluorescence
of a samplesincethe tristimulusfiltersare
placedbetweenthe sampleanddetectorasillustratedin Fig. 8. A com-
puter may alsobe purchasedwith the instrumentto directly measure
color difference in MacAdam units.
A competitive tristimulus colorimeter called the Hunterlab D-25
Color DifferenceMeter (19),* schematically
presentedin Fig. 9, in-
cludesvacuum phototubes,calibratedtristimulusfilters, and a transis-

* HunterAssociates
Laboratory,Inc., 9529LeeHighway,Fairfax,Va. 22080.
658 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

SPECULAR
INSERT
'•'-•----.._••GHT
REFLECTANCE
-----,,,•f
--lt
SOURCE REFLECTANCE
STANDARD
SAMPLE /•< ....
DIRECT LICHT
STOPS

STRAY LIGHT
STOPS
T•NS•SSION
TP•NSMISSION
STANDARD

ZEROING SLIT
- MICROMETRIC SLIT
(VERTICAL ApERATURE)

SPLIT
LENS •'•

ROTATING
"FLICKER"
MmmO• , •, ,,,,•• ,
. TRISTIMULUS and
ABRIDGED
STOP SPE CTROPHOTOME TER
• t FILTERS

PHOTOMULTIPLIER I
I

Figure 8. Block diagram of the I.D.L. Color-Eye

torized color measurementcircuit. The readoutis presentedin X, ¾,Z

tristimulus values as well as L, a, b scales(20) which closelycorrelate
with the Munsell system. The L measureslightnessor value, whereas
the a and b readingsrepresentchromaticity coordinates. A computer
may be purchasedwith the instrument to give direct color difference
measurements in NBS units. The Hunterlab Color Difference Meter
doesnot offer the 10 or 16 wavelengthspectrophotometricpoints offered
in the Color-Eye, which are useful wherenonmetamericmatchesare re-
quired. The Hunter colorimeter may, however, be used in combina-
tion with a Beckman DU or other diffusereflectancespectrophotometer
if nonmetameric matches are required and may be applied to routine
control of pharmaceuticalcolor formulations.
Another commercial instrument, called the Davidson and Hemmend-
inger Colorant Mixture Computer (COMIC), enablesthe coloristto
measurereflectedlight from a sampleof variouswavelengths,usingK/S
 COLOR AND ITS MEASUREMENT (359

COLOR-MEASUREMENT
CIRCUIT

.¸.
¸
¸ a -b

L•

L _ TRISTIMU
LUS
FILTEI•S

CLEAR
METHACRYLATE
LITE PIPE

Figure 9. Block diagram of the Hunterlab D-25 Color Difference Meter

valuesto make a nonmetamericmatch (21). The techniquesinvolvedto

perform thesecalculationswill be discussed.
REFERENCE STANDARDS

Sincemost tristimulusdata are a comparisonof luminosity and chro-

maticity, referencestandardsmust be used as a means of comparison.
An ideal reflectancereferencestandard would diffuselyreflect 100% of
the visible light impinged on its surfaceat all wavelengths. Magnesium
oxide,magnesiumcarbonate,barium sulfate,and white vitrolite have ap-
plication as reflectance standards. A disadvantage with MgO,
MgCOa, and BaSO4 is that they are brittle and yellow with age. The
yellowing processis slowestwith BaSO4. Their advantage, however, is
that these standardshave a diffusereflectanceof about 97-99% in the
visible wavelengths,makinx reflectancereading almost absolute.
660 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

White Vitrolite, usedas an NBS referencestandard,is an opal glass

material with a fire-polishedsurface. The absolutereflectanceis 90%
over most of the visible spectrum, requiring CIE correction factors at
each wavelength. Vitrolite is a permanent standard, is easily cleaned
and excellent for routine colorimetric work.

SAMPLEPREPARATIONTECHNIQUES

Powders

Sample powdersare presentedto most tristimulus colorimetersin a

manner similar to reference standards. A dish with a 4 in. diameter and
•/•-in. depthis sufficientasa sampleholder. The powderisoftenpacked
and smoothedwith a spatula; a glasscoverslipmay be placedover the
sample,provided the referencestandardhas a similar cover.
Tablets

Since the 2- to 4-in. sample viewing area found in most tristimulus

colorimetersis greater than most tablets, it is suggestedthat sample
holdersof • in. or smaller be made, and the tablets measuredindividu-
ally wherehigh precisionis required. The light beam of the instrument
would have to be adjusted here in order to decreasethe beam diameter
hitting the individual tablet.

Suspensions
The reflectanceof opaquesuspensions may be measuredfrom a plas-
tic cylinder with a glassbottom. Caution shouldbe exercisedto be sure
that the precipitate is well suspended,resultingin uniform colorreadings.
Clear Solutions

The color of clear solutions is best measured with transmittance tri-

stimuluscolorimetrysincereflectancemeasurementsoften lead to exces-
sive light scattering. If reflectancemeasurementis required, one may
place a standardizedwhite ceramicbackgroundin the samplecell; the
clear solution transmits the light onto the background which in turn re-
flectsthis light to the photodetector. The thicknessof the solutionmust
be controlled by adjusting the volume of sample used; a 5-ml glass
sampleholder is feasiblefor darker or highly coloredsolutionswhereasa
volume of 10 ml may be usedfor nearly clear and colorlesssolutions. It
should be noted that the light beams travel twice the distance with this
technique,making the reading approximatelytwice the stimulusvalues
obtained with normal reflectancetechniques.
 COLOR AND ITS MEASUREMENT 661

QUANTITATIVECOMPUTATION
OFCOLORANDCOLORFORMULATION
Kubelka-Munk Method

Sincea number of different dye combinationsare often possiblewhen

matching a given standard and the useof many FD&C dyesis becoming
limited, formulators must often resort to color combinationsfor matching
purposes. However, many matchesare metameric and therefore require
spectrophotometricas well as tristimulus analysis. Spectrophotometric
matches, used in combination with tristimulus values of a mixture, tend
to retain the characteristicfeatures of each colorant. An existingprob-
lem with spectrophotometricreflectancecurvesis that these curves do
not directly follow Beer's law since concentrationis not directly linear
with reflectance. Kubelka (22) developednumerousformulasfor cor-
relating reflectancewith concentrationby making scatteringand surface
differencecorrections. It was found that the ratio of light absorptionto
light scattering at a given wavelength is proportional to the concentra-
tion of the dye in the sample. The relationshipshownhere is derived
from the Kubelka-Munk equation.

(• - •)• (•)
Where R = 1.0 at 100% reflectance.
Relationshipof K/S to concentration
K/$ = kc (2)
Where

K = light absorbed
S = light scattered
k = constantof proportionality
C = concentration of colorant

SinceK/S factorsfor eachdye at a particular wavelengthare addi-

tive when mixed together,this principlemay be usedas a basisfor com-
puting the amountsof variousdyesnecessaryto match a given standard.
A sampleequationfor findingK/S of a colorantmixture is:
(K/$),•x. = a(K/S)A + b(K/$)• + c(r/$)c + ...w(K/X)wh•e •se (3)
Where
a = concentration of colorant A
b = concentration of colorant B
c = concentration of colorant C
w = concentration of white base
662 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

This equationwould be usedas follows: The K/S for each dye and ex-
cipient are first obtained by measuringthe diffusereflectancefor a known
concentration of the dye and excipient or white base at a given wave-
length. This R value may be substituted in the Kubelka-Munk equa-
tion (eq. 1), in order to calculatethe K/S values, or these may be ob-
tained directly from tableswhich giveK/S valuesfor reflectancereading
between0 and 100%. The K/S value of the diluent shouldbe subtracted
from each K/S of the dye to obtain the correctedK/S values. For
example,one can calculatethe K/S valuesof three dyes, blue, red, and
yellow,separatelyat onewavelength. The total K/S valuesat that one
wavelengthwill equal the K/S value of a brown, which resultsfrom the
proper concentrationsof the above three dyes. Therefore, if the
value of the sampleis knownat three wavelengthsand the K/S valuesof
the blue, red, and yellow (at any concentration) are also known at the
samewavelength,one may set up three simultaneousequationsto calcu-
late the actual concentrationsof blue, red, and yellow used to make the
match. The Davidson and Hemmendinger COMIC Computer per-
forms these calculationsat 16 wavelengthsbetween 400 and 700 m/• and
alsocalculatestristimulusvaluesfor the purposeof colormatching (21).
WeightedOrdinate Method
This method, mathematically illustrated earlier in this presentation,
is a meansof calculatingtristimulus valuesand chromaticity coordinates
from reflectance data. It should also be stressed that x, y, and • are
tristimulus values resultingfrom the averagevisual observerand are used
as imaginary primaries under specificillumination for tristimulus cal-
culations. The values X, Y, and Z, at any wavelength, correspondto
the magnitude of these primaries needed by the standard observerto
match a color. In addition, chromaticity coordinates,x, y, and z, are
the tristimulus valuesX, Y, and Z expressedas fractionsof their total,
which equal one. These chromaticity coordinatesof the tristimulus
values are then plotted on an x, y chromaticity diagram and the spec-
trum locus is located. The weighted ordinate method is a means of
determining tristimulus valuesand samplelightnesseither from spectro-
photometric reflectancecurvesor directly using tristimulus colorimetry.
Using this method, the relative energy of the light sourceE c (for illumi-
nant C) at a particular wavelength, which is found in published color
tables (23), is multiplied by the tristimulusvalue •c and reflectancevalue
for the samewavelengthbetween400 and 700 m/•, usually 10 m/• apart.
The values are then summedover the total range to yield the X, Y, and
 COLOR AND ITS MEASUREMENT 66;3

100 ' I I I

80-

60-

400 500 540 600 700

•mu

Figure 10. Diffuse reflectancespectra of FD&C Red #3 Dye (100 mg) and light magnesium
carbonate, USP (2.00 g). Key: A, standard dye spectrum: B, new sample lot spectrum

Z tristimulusvalues. Chromaticity coordinatesare then calculatedby

expressingthe tristimulus values as fractions of their total values. The
luminosity or lightness is determined directly from the Y tristimulus
value expressedin per cent.
Although there are other available methodsfor calculating tristimu-
lus values(23), the weightedordinatetechniqueservesas a goodmeans
for deriving color values from diffuse reflectancemeasurements.
I•HARMACEUTICAL APPLICATION

Quality Controlof FDiC Dyes

The purity and quality of pharmaceutical dyes may be controlled, as
illustrated with the following example: A manufacturer purchasesa
newlot of FD&C Red#3 dye. The predetermined
standardspectrumof
this dye compared to a MgCO3 reference standard is shown in Fig. 10
The spectrum for the standard dye is also diluted in a 1:20 ratio with
MgCO3, sinceit is important that the color componentbe highly diluted
with a relatively nonabsorbingdiluent. This ensuresa similar grain size
for the dye and reference standard, causing the reference and sample
scatteringcoefficientsto canceleach other in comparisonmeasurements.
This facilitatesscatteringto be independentof wavelength(24).
664 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

The specificationsfor 100 mg of standard dye diluted with 2.00 g of

MgCO3, (Fig. 10,A) are asfollows' Xm•x= 540 mu; R = 62.8% (K/S
= 0.1102); Tolerance = •- 10%. A new lot of FD&C dye (Fig. 10,B)
givesthe followingspectrophotometricvalues after 100 mg of the dye has
been diluted with 2.00 g of MgCO3 USP' X..... = 540 m•; R = 65.5%
(K/S = 0.0909).
Since K/S values are proportional to concentration,the purity of the
new lot may be found from the calculationsshown below, assigninga
purity of 100% to the standarddye.
K/Ssta. Csta.
_ (4)
K/Xsamrle Csample
Where

K = absorptioncoefficient
S = scatteringcoefficient
C = dye concentrationin %
0.02 oo%
= --; Csamp•e
= 81.6%
0.0909 Csample
The new samplelot shouldbe rejected,sinceit contains18.4•c im-
purities. It should be emphasized that the colorist must check the
linearity of K/S valueswith concentration,and the experimentsshould
preferablybe performedwithin a closeconcentrationrangein order to en-
surebetter accuracy(24).
Tristimulus colorimetersmay also be used to control the color and
lightness of pharmaceutical dyes. This is done by plotting x and y
chromaticity values for a known standard dye and establishingpermis-
sible color differencesfor new lots. The sample lightnessmay be con-
trolled by comparing the sample's Y tristimulus value to that set for
the standard dye. This is a rapid and easyquality control techniquefor
new samplelots of FD&C dyes.

Dye-A djuvantChemisorption
Studies
Several articles have appearedin the literature with respectto color
stability and other changesresulting from the interaction of dyes with
pharmaceuticaladjuvants (25-27). Results of a dye-adjuvant chemi-
sorptionstudy (25) usingdiffusereflectancetechniquesare presentedin
Figs. 11 and 12. The former representsthe results of equilibrating
FD&C Red #3 dye with starchUSP in an aqueousdispersion
medium,
followedby lyophylization(25).
 COLOR AND ITS MEASUREMENT 005

lOO

2•50 •
300 I
350 •
4.00 5JO 600

Figure11. Diffusereflectancespectraof FD&C Red •3 Dye (30 mg) and starch, USP (10.00
g). Key: A, control; B, sample; C, sampleexposedto 184-hourartificiallight at 2000 f.c.
lOO

2hO 2[0 300 3150 4•)0 5•0 6•0 9Jo

Figure 12. Diffuse reflectance spectra of FD&C Red #3 Dye (30 mg) and lactose, USP
(10.00 g). Key: A, control; B, sample; C, sample exposedto 184-hour artificial light at
2000 f.c.

In Fig. 11, ,4 is the spectrophotometricreflectancecurve of the con-

trol, consistingof a triturated dry physicalmixture of the dye and adju-
rant. Curve B showsthis spectrumof the samedye-starchcombination
placedin a suitable container, with 20 ml of distilled water added. This
sampleis then equilibratedfor 24 hours at 30 øC and the material is dried
by lyophilization. Curve C resulted when the equilibrated and dried
sampleis exposedto 184 hours of artificial light at 2000 f.c. using an
Envira-Lite Cabinet.*
Figure 12, ,4, B, and C representa FD&C Red #3-lactosecontrol,
sample,and light-exposedsample,respectively. A comparisonof Figs.
11 and 12 indicatesthat both systemsundergointensespectrophoto-
metric changesduringequilibration,especiallyat the Xm•x
of 540 m•. It
is of interest to note that the dye-starchinteraction is far weaker than the
dye-lactosecomplexsincethere is about 78% fading observedat the
* Thermal Research, Inc., Thermal Road, Iselin, N.J.
666 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS

X.... in the starchsampleas comparedto about4% fadingin the lactose

sample(25).
These studies and others previously conducted on drug-adjuvant
interactions(28) point out the needfor dye incompatibilitytestsin con-
junction with other considerationsin the formulation of pharmaceuti-
cals.

CONCLUSION

The fundamental measurement prindples indicate that the use of

instruments for the control of color will continue to be of great benefit in
the cosmeticand pharmaceuticalindustry. Although future problems
will not be totally solvedwith the aid of instrumentation,the amount of
information gainedfrom thesetechniquesmore than justifiesthe useof
these methods for the control of color.

(Received December 18, 1967)

REFERENCES

<1) Lach, J. L., and Bornstein, M., Diffuse reflectance studies of solid-solid interactions I,
J. Pharm. Sci., 54, 1730 (1965).
<2) Swartz, C. J., and Cooper, J., Colorants for pharmaceuticals,Ibid., 51, 89 (1962).
<3) Lachman, L., Swartz, C. J., Urbanyi, T., and Cooper, J., Color stability of tablet formu-
lations II, Ibid., 49, 165 (1960).
<4) Lachman, L., Weinstein, S., Swartz, C. J., Urbanyi, T., and Cooper, J., Color stability
of tablet formulations III, Ibid., 50, 141 (1961).
<5) Lachman, L., Urbanyi, T., Weinstein, S., Cooper, J., and Swartz, C. J., Color stability
of tablet formulations V, Ibid., 51,321 (1962).
(6) Goodhart, F. W., Kelly, M. A., and Lieberman, H. A., Characterization of tablet colors
obtainable from some certified dyes, Ibid., 54, 1799 (1965).
(7) Everhard, M. E., Dickcius, D. A., and Goodhart, F. W., Spectrophotometric reflectance
method for matching the color of solid dosageforms, Ibid., 53, 173 (1964).
(8) Raft, A.M., Measurement of a color gamut obtained from FD & C colorants,Ibid., 5/i,
380 (1964).
(9) Raft, A. M., Tristimulus color measurements in fading studies and other color changes
of pharmaceuticaldosageforms, J. Soc. CosmeticChemists,18,367 (1967).
(10) Harris, C., Jr., Color Instrumentation Seminar, Instrument Development Laboratories,
Attleboro, Mass., 1966.
(11) Ives, H. E., The transformation of color mixture equationsfrom one systemto another iI,
J. Franklin Inst., 195, 23 (1923).
(12) Guild, J., The colorimetric propertiesof the spectrum,Phil. Trans. Royal Soc.,London,
A230, 149 (1931).
(13) Nickcrson,D., History of the Munscll color systemand its scientificapplication,J. Opt.
Soc. Am., 30, 575 (1940).
(14) Wright, W. D., The Measurementof Colour, 3rd Ed., D. Van Nostrand Co., Inc., Prince-
ton, N.J., 1964.
(15) Committee on Colorimctry, The Scienceof Color,Optical Society of America, Washing-
ton, D. C., 1963.
 COLOR AND ITS MEASUREMENT 667

(16) Judd, D. B., A maxwell triangle yielding uniform ehromatieity scales,J. Opt. Soc. Am.,
25, 24 (1935).
(17) Adams, E. Q., X-Z planes on the 1931 I.C.I. system of colorimetry, Ibid., 32,168 (1942).
(18) MacAdam, D. L., Visual sensitivities to color differences in daylight, Ibid., 32, 247
(1942).
(19) Hunter, R. S., Photoelectric color difference meter, Ibid., 48, 985 (1958).
(20) Judd, D. B., and Wyszeeki, G., Color in BusinessScienceand Industry, 2nd Ed., John
Wiley and Sons, Inc., New York, N.Y., 1963.
(21) Stanziola, R., Practical colour instrumentation for the dyer, Can. Textile J., 1-5 (Sep-
tember 17, 1965).
(22) Kubdka, P., New contributions to the optics of intensely light scattering materials,
J. Opt. Soc. Am., 38, 448, 1067 (1948).
(23) Hardy, A. C., Handbookof Colorimetry,The M.I.T. Press, Cambridge, Mass., 1936.
(24) Lieu, V. T., and Frodyma, M. M., Selection of the optimum concentration range for
reflectance spectrophotometric analysis, Talanta, 13, 1319 (1966).
(25) Bornstein, M., Walsh, J. P. Munden, B. J. and Laeh, J. L., Diffuse reflectance studies
of dye-adjuvant ehemisorption, J. Pharm. Sci., 56, 1410 (1967).
(26) Laehman, L., Kuramoto, R., and Cooper, J., A study of the interaction between qua-
ternary ammonium compounds and several certified dyes, Ibid., 47, 871 (1958).
(27) Scott, M. W., Goudie, A. J., and Huetteman, A. J., Accelerated color loss of certified
dyes in the presenceof nonionic surfaetants, Ibid., 49, 467 (1960).
(28) Bornstein, M., and Lach, J. L., Diffuse reflectance studies of solid-solid interactions II,
Ibid., 55, 1033 (1966).